On the “Wisdom” of Mobs in NCAA Brackets
I’ve always been interested in social dynamics, which is why I’ve found myself so intrigued by prediction markets. I recently came across Andrew McAfee’s post on the “The Good and Bad Kinds of Crowds.” In this post he discusses the use of prediction markets to pick a better NCAA Bracket.
This year I picked two brackets. One with my friends for money, and one friendly one w/ the people over at Abnormal Returns. In the one with my friends I used my standard, “random-ish” method and for the AR bracket used some of the information I was privy to from our markets at yoonew. In using the prediction market data, I didn’t pick solely with the crowd but decided whether I would pick with them or against them. yoonew traders werent so big on MSU, so i picked them. They were big on Villanova, and I picked against them. I picked Missouri with them for a while, but picked Memphis to knock them off. I also picked Duke, which the market was not a believer in at all. Overall I am doing much better in the one using the prediction market data, but even if I get the rest of the picks right, I will be a few points shy of victory. I would think that based on the sophistication of the people in the AR pool – that more than a handful used prediction markets to make their picks, but this is just a guess and would be interested to find out who used what strategy.
I used this strategy to test an idea about Crowds and Mobs. An idea I’ve been thinking about lately is the “wisdom of the mob”. If the crowd becomes reliant on prediction market data, doesn’t it then become a mob? And if so, won’t this mob have the same reflexive properties and “boom-bust” behavior that Soros writes about in “The Alchemy of Finance”, et. al? While a crowd has wisdom, there is very little “wisdom” in a mob. Therefore it is much more useful to be able to differentiate from a random crowd and mob and know when to pick with the crowd and against the mob.
As many people in the prediction market space look for ways to monetize the data, too many are looking at selling the data as if it were wisdom, but it is only information. I think the company who can use prediction markets is to be able to discern whether the information one gets comes from a crowd or a mob is infinitely times greater than just producing data like everyone else.
In the pool, I believed that Lousiville and Pitt yoonew voters had become a mob, and rightfully so, but overlooked the “crowd” nature of the Villanova. From a Monday-Morning Quarterback view, I can now see how powerful (and feasible) it is to discern between the two.
This is why I think integration with Twitter is powerful. A stream of random “tweets” could be viewed as a crowd as each tweet itself stands alone. While on the other hand a series and network of @ replies is a conversation that can take the form of a mob. Therefore the more links a piece of information has, the less valuable it becomes in the market, but the more random hits it has the more valuable it becomes. I curious to hear any thoughts out there.